4 More Years? 😆 #MyXAnniversary

AYO! We have a 2-year postdoc fellowship available immediately to work on computer vision / VR experiments with human infants. Co-advised by myself, Aliza Sloan, and Scott Kelso at Florida Atlantic University. Ideal candidate is a biologist with programming/modeling chops. Pls circulate!

Philip Ball says that all animals, down to single cells, have goals and agency, but not necessarily consciousness

Very important and grounded takes by Machine Learning Street Talk and Philip Ball - Highly recommend a listen.

youtu.be/n6nxUiqiz9I

TyfoodsForThought Captain Pleasure, Andrés Gómez Emilsson Yep! I am working on multi-agent LLM systems to automate theory validation in neuroscience. You have many agents who look at a theory, generate and run experiments to test it, and generate a report on all of it. You can sign up for progress updates here: forms.gle/GUM5hRDztxQZx9…

I absolutely agree

Giulia Biffi No, but you get used to it. Everyone is fighting their own battle - their rude comments are about them & their pressures, not you. No one has a crystal ball, they can't tell good ideas from genius ideas, and they have to pick the top 8%. It's a numbers game; improve + repeat...

Paper accepted! We show how avoiding predators and tracking resources in space changes pattern formation in ecological networks-of-networks (preprint version is the accepted one): arxiv.org/abs/2403.13098

there's nothing more physically painful than not knowing why a bug exists. but there is a point where i must close the text editor. it will honestly be faster if I take a break. BUT EVEN SO. I CANNOT CLOSE THE TEXT EDITOR. I MUST KNOW

'If our brains were simple enough for us to understand them, we'd be so simple that we couldn't.'

The future that I live for 🤩😂
Very cool

Very nice video on Von Neumann by The Last Theory

I recently argued that we need to complete Von Neumann's dream of programmable self-replicators. I think the reason that 'Von Neumann' isn't a household name is because he died before finish *his* big work

youtu.be/17mNnWn2AsM?si…

The second, and follow-up to 'Popper Primer' posted yesterday: 'Deutsch Digest' where I give my interpretation of where David Deutsch has built on the work of Karl Popper. All in under 6 mins!

Special edit and captioned for X

Errors my own.

'Upgrade your brain'!

The amount of free will you have a particular time depends on the amount of knowledge you have about your possible choices at that time. In this sense, knowledge makes you more free.

On March 29, I'll be joining Active Inference Institute to talk about my recent work on generalizing the partial information decomposition to the Kullback-Leibler divergence, and discussing what this might mean for active inference and the free-energy principle.
youtube.com/watch?v=Q7_LwM…

'Words and drugs have the same mechanism of action' - Fabrizio Benedetti

Understanding OpenAI'S Sora a little better (image patches as tokens, adjacent frames etc.) - suggests no physics 'model' in any normal sense of the word, just mimicry from zillions of sources - very good, probabilistic mimicry (like GPT), but nothing more...

The misconception that knowledge needs authority to be genuine or reliable dates back to antiquity, and it still prevails. To this day, most courses in the philosophy of knowledge teach that knowledge is some form of justified, true belief, where ‘justified’ means designated as…

The future that I live for 🤩😂
Very cool

Have you ever done a dense grid search over neural network hyperparameters? Like a *really dense* grid search? It looks like this (!!). Blueish colors correspond to hyperparameters for which training converges, redish colors to hyperparameters for which training diverges.